#93 – Beyond the Code: Jeremiah Owyang on the New Era of AI

BOUNDARYLESS CONVERSATIONS PODCAST - EPISODE 93

 placeholder
BOUNDARYLESS CONVERSATIONS PODCAST - EPISODE 93

#93 – Beyond the Code: Jeremiah Owyang on the New Era of AI

Jeremiah Owyang, a veteran with a vantage point of Silicon Valley and current General Partner at Blitzscaling Ventures joins us for the second time on the podcast. While he is now working on identifying and nurturing AI startups, this episode reflects his art of captivating audiences. He takes us through the exponential growth of AI startups, its implications on creativity, health, wellness, and productivity, the ethical considerations surrounding its adoption, and the potential of AI to democratize professional skills. We even touch upon what’s next with AI in this thought-provoking discussion, so tune in and get listening into the future.

 

Youtube video for this podcast is linked here.

 

Podcast Notes

In this episode, Owyang demystifies complex technologies – a skill that has made him a sought-after speaker, capable of creating context and setting standards for the adoption of advanced technologies in Business, globally.

Jeremiah takes us on a journey from the inception of predictive AI to the emergence of ChatGPT the moment it was made ubiquitous. This episode is special since we go beyond the technology – through the socio-economic impacts of AI – with a bonus live demo of scenario-based voice interaction with an AI.

This episode is a deep dive into the future of AI and the sheer power of its technology from one of Silicon Valley’s most influential voices. Don’t miss out.

 

 

Key highlights

  •  AI’s role in reshaping creativity and digital content, with digital influencers.
  • Democratization of AI technology beyond the Silicon Valley – its global adoption.
  • Addressing the balance between technological advancement and the preservation of human values and rights, with Ethical AI.
  • Consumer interaction with AI, and what the future will hold – mobile-first or wearables?
  • A step towards artificial general intelligence (AGI) – The emergence of AI agents as autonomous systems enhancing personal and professional productivity.
  • The role of AI in health, creating accessibility, affordability, and overall improvement in general wellness
  • Harnessing AI for Global Betterment – elevating health and education, paralleling the Industrial Revolution’s transformation; akin to trading human cognitive efforts for computational prowess.

 

This podcast is also available on Apple PodcastsSpotifyGoogle PodcastsSoundcloud and other podcast streaming platforms.

 

Topics (chapters):

(00:00) Intro – Beyond the Code: Jeremiah Owyang on the New Era of AI 

(02:29) Picture of the white-hot core of AI

(12:27) AI’s Creativity and Human Creativity 

(18:51) Monetizing the future of AI

(28:18) Authenticity of AI 

(39:32) AI and it’s human implications

(43:39) Breadcrumbs and Suggestions

 

To find out more about his work:

 

Other references and mentions:

 

Breadcrumbs and Suggestions: 

 

Recorded on 11th January 2024.

 

Get in touch with Boundaryless:

Find out more about the show and the research at Boundaryless at https://boundaryless.io/resources/podcast

 

Music:

Music from Liosound / Walter Mobilio. Find his portfolio here: https://blss.io/Podcast-Music

Transcript

Simone Cicero (00:10.128)

Hello everybody and welcome back to the Boundaryless Conversations podcast. On this podcast, we meet with pioneers, thinkers and doers and we talk about the future of business models, organizations, markets and society in this rapidly changing world. Today I am here with my usual co-host Shruthi from India today. Am I right?

 

Shruthi Prakash (00:32.799)

Yeah, I’m here from India. Thanks, Simone.

 

Simone Cicero (00:35.816)

Thank you so much, Shruthi, for joining. And I am also with an incredible and I’m incredibly excited to have a returning guest, a friend with us, Jeremiah Owyang, which is who is a visionary from the heart of Silicon Valley. I would say Jeremiah is now a GP at Blitzscaling Ventures, where his exceptional talent is now focused on identifying and investing into AI startups that show, let’s say, promise for rapid growth. Jeremiah’s work doesn’t end there. 

 

He’s a captivating speaker who has been demystifying advanced concepts for companies for ages now, and now he’s doing it with AI. Whether he’s broadcasting from this tiny, high stream studio or being in person energizing crowds of business events or other live conferences is always giving enlightening insights. 

 

It’s worth mentioning that I know Jeremiah since more than a decade, that we were both early pioneers in the collaborative economy. And I remember fondly the days where I was a program director at the Wisher Fest, and I had you as a speaker. 

 

So Jeremiah, your journey in tech has been remarkable. I really hope you can enlighten us on AI today. And welcome and thank you for being with us.

 

Jeremiah Owyang (01:40.588)

Yes. Thank you for having me. What an honor to be here and our friendship sustains over time and distance.

 

Simone Cicero (02:09.404)

Thank you so much. So let’s start without further ado and jump into the topics. So I think, you know, you are really inside the core, the white-hot core of the AI, let’s say research, the development and everything that is happening in Silicon Valley. So like with other trends, you have been pioneering, you have been early on this AI trend. 

 

And so how do we start better than asking you to kind of make a picture of what’s happening? How you have been putting yourself in the center of it and what threads you can put together for us to understand what’s going on?

 

Jeremiah Owyang (02:56.014)

Thank you. Well, obviously AI is not new. It’s been around for many decades, but that trend has been known as predictive AI or machine learning. When it came to November 30th, 2022, we saw that birth of GPT, which caught the world by storm. Within five days, they had a million users, which is one of the fastest growth rates ever of any consumer app. And people could actually talk. It was a real experiment that anybody could use. And there’s a hundred million users plus on that under a year. So that really took the world by storm of generative AI. 

 

So it’s important to distinguish the different lineages here. So when I started to track on a website called thereisanAIforthat.com, of course, in March last year, 2023, 2022, there was actually 3000 AI startups. Today in January, there’s 11,000 startups. That’s 1,000 new AI projects being created per month. Most of them, all different countries are representative in creating that. It’s not limited to Silicon Valley or to London or Paris. It’s happening all over the globe. The cost of entry is, it’s really easy to get into this. 

 

Now with that said, I am in Silicon Valley this is my fifth tech wave. And Simone, I’ve never seen a wave this big. This is something different. 

 

Just to help paint that picture, there’s about one third of the billboards on Highway 101 are around AI. Of course, this is a very nerdy area, so that’s not unheard of to see that many tech billboards, but to be all focused on one trend is amazing. The other thing to note is that every big tech company, all the tech giants, they’re all now AI companies. And of course, companies like Google are very afraid. They could potentially lose market share as people turn to AI for questions and answers, that’s their main revenue. The other thing to note is there’s a tremendous amount of AI founders and startups, all in Silicon Valley and in San Francisco in particular, that are creating these startups and they’re launching these things. One final stat that I’ll share and then I’ll turn it back over to you for your comments is just to explain the growth and the excitement and the ecosystem. 

 

There are 100 physical in-person AI events in San Francisco per month. 100 per month that’s three every day on a weekday two on Saturday and two on Sundays – which can often be hackathons. Most of the foundational models the biggest ones are in San Francisco. Of course, there are some around the globe like Mistral in Paris and UAE has Falcon. So, and early other models came from Toronto and London. So we certainly know there’s a global ecosystem, but the real energy is absolutely in San Francisco. So that’s a quick quote, reporter on the ground view that I wanted to share with you and your listeners.

 

Simone Cicero (06:05.804)

Thank you so much. I mean, I think you said something very, very on point when you said, this time is different, right? So my question for you, again, it’s kind of bouncing back to you based on the experience you had in the last few months. We have been meeting so many founders, you have been meeting so many people that have been active on the space. Sometimes it feels like it’s just, you know, it’s just in fashion to be AI something, right?

 

But on the other hand, I believe it’s really a deep structural shift in how we think about products, how we think about organizations and so on. So maybe you can help us in a further reflection to explain why it’s really different. And it’s not just a fashion element, it’s really a deep structural element that is impacting the way we think about products and everything else. So what’s your point of view on this?

 

Jeremiah Owyang (06:59.918)

So I was involved in the Web3 blockchain industry and it was very difficult to, you know, explain what the use case is and really the purpose. And it was very challenging for users to use. There were so many steps. I couldn’t get my mom to use it. Of course, I believe those technologies will sustain over time, but it’s just a longer adoption process. Now you and I were both in the collaborative/ sharing economy and the use case was very clear. Do something faster, cheaper or more local or beneficial to the planet and sustainability is very clear, right? And that’s why some of those dominant players still exist today. It’s very obvious. 

 

Now, AI is akin to more of the sharing economy. When you use it the first time, you’re like, oh, my life is easier. My life is better. I can get information faster. I can create new things quicker. And so the instant ability to be more productive is the main use case to save time. So it’s, it’s very clear to people to use. 

 

Just editing documents or grammar check or getting new information or just to create some new images. Now, of course, this is not without its flaws, which we should absolutely address. You know, there’s issues with where did the content originate from? Is it being ethical? Are the people being compensated who created the original data corpress? Is it creating misinformation? Those are all real things that are being addressed by the industry. So it’s not all roses. You know, every garden has bugs and thorns as well.

 

Simone Cicero (08:30.241)

So you spoke about, just to continue a little bit on this use case evidence that you’re talking about, you spoke about productivity, right? And productivity sounds like a very enterprise-ish benefit, right? So we think about enterprise software, or it can be also, I would say, for professionals or freelancers and so on, that you can be more productive with AI. That’s clear. I do my research with AI.

 

I use the AI help now in blogging and social media and whatever. But maybe can you just, among the startups you are considering, what are other startups that are using generative AI? Because we’re talking about, massively talking about generative AI here. Is there anything else that generative AI is already showing potential use cases for – beyond productivity, right? So what’s your feeling in terms of any other trends that are emerging?

 

Jeremiah Owyang (09:27.861)

Oh yeah.

 

For sure. And I’d be so curious from somebody who’s lived in Italy, the old world, the center of creativity, right? For centuries on this take. So we’re seeing AI boost creativity. We’re seeing AI create images. We’re seeing it create videos. And some of these images right now are creating new digital influencers like on Instagram. And you can’t tell, you cannot tell they’re AI created. It’s actually impossible to tell.

 

And part of the reason is because most of the human influencers are using post-production edits and digital filters themselves. So there’s like this meeting in the middle. We’re also seeing people create new videos. There’s some new startups that are emerging. The name escapes me at the moment. Pika, yeah, Pika, which you can create videos just from a few text prompts and the videos look like standard animation. You would see from a movie production house. 

 

And of course we’re seeing video production. So we can, everybody can now become a creator and create new forms of poems, stories, images, photos, digital illustrations, art, video. And of course that will bleed into the next phase, which is multiverse and metaverse and virtual worlds and augmented reality, which has been, it’s going to be released by Apple this next few weeks. This week is CES at Las Vegas.

 

So all these trends are coalescing together, right? An AI doesn’t exist just on your small screen, on your mobile device. It’s going to spread into the TV. It’s going to spread into a virtual reality and then AR right now, both of you have glasses on your head. Future versions of glasses could have a heads up display that could give you more information around the world. So you can either create or be more productive or figure out how to relax. But so I want to hear your take on that creativity and AI.

 

Simone Cicero (11:22.136)

Yeah, I mean, so for what I perceive from you, you’re talking about mainly seeing use cases related to productivity on one side, creativity on the other side. And I see that, you know, professionals can use it as a sparring partner in their creative endeavor. You spoke about entertainment and gaming and movies. You know, I’m fascinated by seeing what’s coming up in AI-based video. It’s really mind-blowing sometimes.

 

So can we say productivity, creativity, entertainment, anything else that comes to your mind?

 

Jeremiah Owyang (11:56.854)

Yes. An important one would be health and wellness. And right now, there’s a real big issue in the United States. So the health span and lifespan are decreasing. And it’s part of that was due to COVID. But Americans, as you know, are woefully out of shape and unhealthy. But it’s that behavior and is spreading to other countries, UK, China, Mexico.

 

Like people are getting out of shape and also there’s the highest levels of anxiety and depression ever recorded in the USA at this time. So people are turning to AI to help them. Some of the, and I can show you a demo later, like there’s one called Pi where people talk to that one or there’s other ones for mental health or people using these for meditation or they might use it for fitness to come up with a personal plan and then you can connect it to your wearables to track your fitness and are you moving during the day. So health – life, health span, mental wellness, relaxation, sleep, AI can assist and also detract from those as well. So this is a tool that can be used for that.

 

Simone Cicero (13:08.902)

Mm-hmm. Shruthi I know that you have a couple of questions lingering there?

 

Shruthi Prakash (13:09.983)

Yeah. I was, I was just going to jump in. Okay. So, I mean, this is regarding the previous point, but I’ll, I anyway want to hear from you on it. So does let’s say AI’s creativity one come at the cost of human creativity. So is that sort of a trend that you’re seeing is that it’s either or, and how is the reception for all of this? Right. Essentially like how is consumer behavior changing in order to receive this kind of AI driven creativity.

 

Jeremiah Owyang (13:43.43)

Ooh, good, great questions. And I just wanna comment, I mean, this is a really cool global conversation. We’ve got somebody in Europe, Asia, and I’m in North America. So this is like really fascinating and all in real time. Okay, so how many people are actually creative? Right? That’s a question. What do you think?

 

Shruthi Prakash (14:06.027)

I mean, I think all of us are, it’s probably degrees of difference.

 

Jeremiah Owyang (14:08.491)

Hey and how many people have had formal training to be creative.

 

Shruthi Prakash (14:15.699)

I mean in an Asian context, extremely low, I would say.

 

Jeremiah Owyang (14:18.97)

Oh, yeah. So now we have the tools, the AI tools where everybody, an amateur, a quote, creative amateur can level up to be a prosumer, which is in between a professional and a consumer. And those prosumers can level up to become like professionals. And the professionals, well, they better step their game up because everybody else is leveling up too. And of course they will something new will happen. And so this is this example of everybody gets to step up a level.

 

How were people reacting to this? That’s a deep question. So the professional class at first was very, very upset with this. And Simone, you know from your history in Europe that people were upset during the Renaissance period or when Impressionism came around. Every new movement, people were upset with it. Even people were upset with the Eiffel Tower when it was produced. So people were upset with generative art. But last week, Getty Images announced that they will now sell generative AI art.

 

That only took them less than a year to turn around completely and figure out how to make this part of the culture and to become common. Of course, their art is derived from their official art, which they have in their collections and they have IP ownership of. So that gives them power from their own business model. So this is an opportunity for all the content to level up. Now, are we ready for synthetic content created by AI to surround us at all times? Honestly, it could be a little overwhelming. It is overwhelming. I see it on Instagram. It’s like all this synthetic art. It is overwhelming. So every time we have one reaction, we will go back to the other direction. So there will be a premium on human. And so when things move too far to synthetic, we will shift back. And we saw this in the sharing economy. It was a reaction to over-consumerism. It was a reaction to too many cars in the US, right?

 

is a reaction to people being stuck in the same old building. They wanted to be an Airbnb and travel the world and not be in a hotel. So if we have too much synthetic content, we’ll see people go back to museums to see 1000 year old pieces. If there’s too much synthetic content, people will go back into nature and have a nature hike, which is a much higher resolution than 4K. So that’s my response to that. Those couple of permutations on that thought that you that you gave me. Thank you.

 

Simone Cicero (16:44.508)

Are you seeing, I mean, this is a connection I’m making with the sharing economy and the collaborative economy as well. When we have seen these new models coming up, you know, I think there was a phase of this adoption that was very characterized by the question of trust, right? Because we were used to, let’s say, being a consumer in the traditional environment.

 

And suddenly we were facing peer-to-peer relationships. We didn’t know how to trust the others. And we were having no customer protection and so on. So are you seeing similar concerns for interacting with AIs in terms of, can I trust this AI? Can I give, I don’t know, any trust issue related to the emergence of this AI mediated? Models in startups. What is your perception of how consumers are approaching these new services in this way?

 

Jeremiah Owyang (17:47.15)

There’s absolute mistrust of the hallucinations created by the large language models like OpenAI, which in the past have sometimes made up things. Even my own biography was slightly incorrect because it’s just trying to fill in what it is, it was assuming would fit naturally into the next word order. That’s really what it does is reassemble the word salad, right? However, that’s being solved at a rapid pace. There’s a new technology trend that has really took the industry by storm come November, 2 months ago, called retrieval augmented generation, otherwise known as RAG. And what that means is the large language models like OpenAI can access other data sources to get richer information. As an example, if you use OpenAI and ask it to shop for like cameras, it will go to eBay and Amazon and to Best Buy and other stores to go get prices. And that’s happening now. It won’t just make up a price.

 

and then it cites a small source. If you use the fast growing new search tool called perplexity.ai, have you guys tried this perplexity? It will show citations at the top of where it’s getting its information and it will summarize the answers from your query right on the page with small little footnotes so you know where it’s getting its information. So by the end of this year 2024, hallucination should be gone on these types of tools. However, there’s a new problem. 

 

The new problem is there’s multiple democratic elections happening around now, and there’s the big one in the US happening this year. And we’re expecting that foreign state actors who are in opposition, or frankly, Americans themselves, will be creating false, fake, deep fakes, making public officials or celebrities say things that weren’t true; or to stitch together multiple clips which were true, but really lend and change the context. So people, bad people, good people, will be using AI to distort truth and reality, and it will appear real. So some of that will be fake content, some of it will be manipulated content, and that is starting already.

 

Simone Cicero (20:10.068)

Yeah, I mean, Shruthi unless you want to jump in on this, I was thinking that, I mean, this connects with a couple of, you know, I’m thinking about a couple of things I would like to discuss with you, but certainly this is one. So sometimes I have a feeling that AI is accelerating, right? It’s accelerating also technological development. You have this feeling that we’re running faster, since we invented these new large language models. It really felt like an acceleration, right? And at the same time, as you said, so there are deep fakes and these things are creating massive challenges from a perspective of social cohesion and things like that. So I feel like it’s like we’re running faster on a car that is falling apart a little bit as we go. And I mean, we see this reflected in the challenges that AI is making for policymakers, for example. 

 

This is a massively deflectionary technology, because it’s really making work more commoditized, and it’s going to create a lot of challenges for policies. So from a VC perspective, you’re looking into making money with this. So what is your perception on thinking about the world we will have once AI is fully deployed and how to make money in that world through AI. 

 

So what is your VC investment thesis and monetization thesis on these technologies on a wider perspective?

 

Jeremiah Owyang (21:47.97)

Sure. Understood. So venture capital invests in promising startups that are likely to grow and become the dominant players in the space. And that is an existing business model that’s been around for decades. And it’s quite common, of course, for tech companies to work with VCs because they help them with guidance or to make connections or to set up events or connect them to customers. And so it’s an exciting role for me to participate.

 

So the way that obviously VC works is you invest and then after several years, which could be up to sometimes 15 years, you would see an exit or return in that investment. And that’s how the model works. So typically in some cases, VCs would then use that money to reinvest into the market in other ways. And so it continued forward the next iteration, the next generation of tech. As you know, the tech cycles seem to come in like five to 10 year waves. 

 

And so VC funds tend to match that wave. So that’s honestly how we would make money in doing that is investing in recouping returns later. By the way, not all the startups, most of the startups will not make it. So you actually have to spread the bets around. Typically it’s 5 to 10% of startups in tech will actually be have any return of note.

 

Simone Cicero (23:10.732)

So at the moment I can perceive that since this is more like a very dynamic landscape, you will tend to invest more into radically breakthroughs, like breakthroughs in productivity, breakthroughs in creativity, breakthroughs in content creation and things like that. So do I perceive well that business models are not clear yet. It’s not really clear how do you monetize these technologies fully at the moment, am I right?

 

Jeremiah Owyang (23:42.134)

Oh, okay. No, I think it is clear. So, okay, so it depends on who we’re talking about for business models. So for the startups, the business models right now tend to be SaaS-based. So companies are paying a monthly, which could include if it’s like a B2B, you access to an API or tokens, then obviously you would pay per usage. So that’s pretty common already, like a traditional cloud type of thing.

 

What we will see in the future and it’s starting to come around would be advertising as a revenue model. That will come. In just a few months, I expect to see, “sponsored sentences” in some of the LLMs. Would you ask it a question and it renders a conclusion or a summary? And there might be an ad in there, a sponsored sentence. Like for example, what is the best midsize car to buy? And it might say Honda, BMW, Mercedes. And this one is sponsored by you know, pick a brand. So those are the main business models I see at the moment.

 

Simone Cicero (24:45.484)

I know that, Shruthi, you have a question, but I would like to complete this reflection on business model. So you said SAS. So I guess that SAS is most applicable to the productivity element. Then you said tokens, which is about the creativity generation use cases. And you said also these fantastic ideas around advertising. And basically, I guess it’s about the let’s say putting yourself as an advertiser in these conversations that exist between liking games or things like that. And just a quick question, what is your idea on AI based hardware?

 

Jeremiah Owyang (25:25.73)

So I’m not an expert in that, but I can comment at industry level. So right now, Nvidia is the hot company that is creating the hardware. If you’re talking about chips as being the main funnel.

 

Simone Cicero (25:35.584)

I was talking more about the tabs and the pins and the wearable element, but you can make a broader point. Yes.

 

Jeremiah Owyang (25:39.946)

Oh, got it. Humane. Yeah, gotcha. Humane. Rabbit. Apple Watch Vision Pro, which is coming. Any smartwatch and frankly, this phone that I’m holding up. Right. OK, got you.

 

Simone Cicero (25:52.795)

Yeah.

 

Jeremiah Owyang (25:56.318)

So for most of the world, they are mobile first or mobile only. And so that’s how they would interact with AI. In fact, I can show you a demo of some powerful ones later. It’s audio-based. So yeah, for sure, that’s the case. I’m not convinced people are going to be wearing those humane pins and walking around. That honestly seems like an edge case that may not become normalized for a long time. Don’t forget, there’s glasses. So Meta has the Ray-Ban glasses that have cameras and that can do other things, which eventually have other capabilities. 

 

So I think it’s mostly gonna be on the mobile phone for now, but over time, the mobile phone will disappear and we’ll start to use these wearables as our main interface. One of the powerful things of these AI technologies is that their voice capability is very good and their microphone capabilities are getting great. And putting a small camera or letting your camera be exposed or your GPS location already gives you a lot of context. So the AI kind of knows what you’re doing.

 

Simone Cicero (26:53.052)

So we’re really getting into these Her-like situations, where you speak with the AIs as a background to your daily routines.

 

Jeremiah Owyang (27:00.446)

Yeah, I mean, have you tried PI?

 

Jeremiah Owyang (27:06.658)

I think we should do it. I think we should try it right now because it is very much like her. Could we try it? It’s a voice.

 

Simone Cicero (27:07.13)

No I didn’t.

 

Simone Cicero (27:14.516)

Ah no, I mean yes, sorry, you meant, yes I tried, but if you want to show it, why not?

 

Jeremiah Owyang (27:19.522)

Okay. Let’s do it. Okay. So Pi is a free app. Um, it’s made by inflection AI and I, we have financial ties to inflection AI because they’re co-founders. One of our advisors, Reid Hoffman. All right. So PI is an app. And if you can see the video, it’s just on my phone and I’m going to turn it on.

So, hey, Pi, I’m with my new friends. Tell me what you do in just one sentence. 

 

Conversation with Pi

Pi Absolutely, Jeremiah. I’m here and ready to answer any questions you may have. Go ahead and start fresh. 

 

Pi – I am an AI powered language model that provides personalized assistance and conversation through natural language interactions. 

 

Jeremiah – Hey Pi, I’m feeling really sad. A car ran over my dog. What could I do?

 

Pi – I’m so sorry to hear that, Jeremiah. Losing a beloved pet is a devastating experience. The shock and grief you’re feeling are completely understandable. I’m here to listen, and I’m happy to help you through this difficult time. In the immediate aftermath, it might be helpful to allow yourself some time to process what happened and grieve for your loss. Reach out to friends and family for support if you feel comfortable doing so. You could also consider writing down your memories of your dog or creating a memorial to celebrate their life. There are also grief support groups and counselors available to help you through this difficult time. Remember that it’s okay to take as much time as you need to process your emotions and to seek professional help if you feel like you need it.

 

Jeremiah Owyang (30:06.53)

All right, was that the first time you had an AI on your podcast?

 

Shruthi Prakash (30:06.93)

I think so, yeah.

 

Simone Cicero (30:07.256)

Yes.

 

Jeremiah Owyang (30:09.002)

Alright, what did you observe? What did you hear? What did you observe?

 

Shruthi Prakash (30:13.687)

I was thinking it sounds like some black mirror episode that came to life where we’re falling in love with a mobile device. That’s what I was thinking about.

 

Simone Cicero (30:24.632)

Yeah, I mean, I was surprised by the actually again, in terms of tone and type of interaction is really, really becoming natural. So it’s really impressive how fast this technology has progressed in just a few months. And it’s like, I mean, I can think that in six months, you’re really going to feel like talking to a human.

 

Shruthi Prakash (30:45.932)

Mm-hmm.

 

Jeremiah Owyang (30:46.498)

So if you noticed I was hands-free, I mean, I was holding the phone up, but I could just have the microphone on and have a dialogue. And sometimes it engages in dialogue. Like I was expecting it to say, do you wanna talk about your loss? Sometimes it does that. Like how do you feel right now? So like the two-way dialogue, and notice like the tone changed when I brought up something somber. What did you hear?

 

Shruthi Prakash (30:59.544)

Mm-hmm.

 

Simone Cicero (31:00.326)

Mm-hmm.

 

Simone Cicero (31:08.962)

Yeah, yeah, absolutely. That was, again, the tone is the thing that surprised me the most.

 

Jeremiah Owyang (31:16.654)

Okay, so that and the voice, the last thing on this, that voice sounds a lot like Scarlett Johansson from Her.

 

Shruthi Prakash (31:19.351)

Yeah.

 

Simone Cicero (31:24.685)

Yes

 

Shruthi Prakash (31:28.164)

Okay. Yeah. Yeah, I was just, no, I was just thinking about all of this content that’s being produced, right? And so what happens, let’s say as a VC maybe, or generally being in the system as well, right? How do you assess consideration on like manipulations in these content or accountability for the information that’s being said and overall, let’s say your idea on the ethics of it, right? Like how do you think that’s going to progress as we go forward?

 

Jeremiah Owyang (32:04.514)

Yes, important question. So obviously we’re not going to invest in any company that’s unethical or going against the grain of humanity. That makes no sense. It doesn’t make sense financially. It doesn’t make sense for me to sleep at night. It doesn’t make sense for my children. So I don’t want to do that. Obviously every founder that we work with is going to need to follow the laws in the countries that they’re operating in. However, it’s important to note that AI is going into uncharted territories, you know, “Uber and Airbnb were illegal when they started off.” Right? 

 

And Google’s been sued for things that they’ve done and micro loans and blockchain is quote illegal, even though there’s a value in help in those technologies. Of course, these things are not always used in all of them. They’re not always used in an ethical way, but the rules and the laws and the social norms will have to mold. The laws do not reflect what’s happening in AI. 

 

They don’t. They frankly can’t keep up with the changes that are happening. There’s no way they can keep up. And with most governments, well, I’m sorry, not most, with some governments, the leadership team changes over every four to eight years, several years. There’s a few governments where it’s lifetime, right? But they just can’t keep up. They’re not even in office long enough to understand and make the changes. So thank you for that important question.

 

Simone Cicero (33:07.534)

Mm-hmm.

 

Simone Cicero (33:29.38)

I feel like it’s a bit different than in the past, right? Because when we spoke about, for example, Airbnb being illegal, or actually maybe not really Airbnb being illegal, but the way it was used, right? You know, from people without any license or people that didn’t track their payments and so on. In this case, I think there are some more important problems to address that are structural to how do we build, how did we build the LLMs, right? Basically using knowledge produced of which the LLM owners don’t have ownership. You know, so I think it’s some structural problem that we need to figure out and we are already seeing the NIT, you know, suing OpenAI and so on. But I think before we, you know, something that I wanted to discuss with you, because you had a very strong connoisseur of consumer behaviors as well. And I think you also in the preparation conversation we had, you spoke about agents that I think you believe it’s a very important aspect of this AI revolution. And indeed it’s moving from talking to AIs into letting AIs do the work and coming back to us, more like something like, you know, an assistant or a butler or something like that. 

 

So AI agents, AI agents, I think, one question that I have, the biggest question I have with AI agencies, first of all, trying to understand through you, what are they really capable of? And I think they are really becoming smart and can do actions much more easily now. But on the other side, I think it would be interesting to explore from your perspective, what are the security concerns. Because I see that when we talk about agents, most often people have this concern about having to supervise the actions that the AI do, the AI is performing. So essentially controlling that they do not do wrong things. So for example, buy wrong stuff or order things that they don’t need or maybe unadvertably execute some tasks that have consequences in the real world for the user for which the user also has to supervise. So what is your perspective about agents? How are they evolving and these security concerns?

 

Jeremiah Owyang (35:53.71)

Great questions, a group of questions there. Okay, so let’s start with a definition. So right now, LLMs like OpenAI are more like assistants that you would go to and ask it for help. Another version that we’re seeing emerge quickly is called a Co-Pilot, where the AI is integrated into your existing software. That could be personal software that you use for fitness that guides you in your fitness journey, or it could be in Microsoft 365. Literally, it’s gonna be called Co-Pilot that sits with you in Microsoft Word or your actual Teams meetings.

 

and takes notes for you or schedules for you, right? So there’s a co-pilot that’s like with you. The third type, which is now emerging, is called AI agents. And there’s no clear definition on AI agent, but essentially an AI agent has the following characteristics. One, it’s autonomous. It operates with little or no human supervision. And number two, it perceives what’s happening around it which it helps to guide it and make decisions. So it is like the baby step towards artificial general intelligence, AGI, which is basically the goal here, where AI would be equivalent to human intelligence, which by the way is probably gonna happen in our lifetimes. All right, so an AI agent, there’s around 300 startups that I know of. In fact, my next event in San Francisco called Llama Lounge, the focus will be on AI agents, and 10 different AI agent startups will be on the demo floor like an expo hall, and they will have demo tables. 

 

Of course, this helps me to understand who we should invest in and the trends and meet the founders. So I’m already interfacing and interacting with the AI agent space already. And so that’s starting to really grow. So AI agents, so let me just tell you about some of the AI agents, and then I will address your question on security. So one of them is called Agent Lunar, which is from UK. They’re flying in digital workers for small business. So instead of hiring people, these things would do that; a personal shopper agent called Claros. And I’ll give you one more EvaBot, which is AI research assistant in sales that will help sales teams learn more about their customers. And it’ll go out there and do all the research. Normally you would hire somebody junior to do that. So in many of these cases, agents will be doing the low level tasks, repetitive tasks that have already been done by an entry level worker. So most entry level worker tasks within the next few years will be automated. Sorry, I’m sorry. That’s the truth. And so we better prepare ourselves in society and culture for that to happen. So those are what AI agents are. 

 

Now, to answer your question around privacy and security. Well, it only has access to the information that you allow it to. So if you are concerned about your bank account, then don’t give it access to that. But it can certainly go around and use the existing public web. Like we see them doing shopping for people or booking flights. Or if you give it access to your calendar, that’s a common one that I’m seeing. It’ll schedule calls for you and schedule meetings. It’ll look at everybody’s calendar. Right now, it’s very manual with Calendly to do that, you have to both have to pick out what are your free dates and all those things, but it should be able to figure that out by context. Specifically, the three of us are in three radically different time zones. So it should be able to do those things on our behalf. 

 

The AI agent that I expect to see by the end of this year, I was asked by Fast Company to give a prediction for this year, 2024. And so by the end of this year, I predict that we will see AI agents that will help us solve our email inboxes.

 

Right now, all of our email inboxes are a disaster. And it’s so much anxiety and stress to think about all those unread emails. And of course, as soon as you respond to all those emails, guess what happens? More emails come back. Yeah. So basically, I’m expecting AI agents to read emails, prioritize emails, summarize emails, and then recommend its responses. And then after we see the recommended responses get better and better, start to say yes, yes. And then eventually you give it the autonomy to respond on your behalf. So that will happen over time where your AI agent will be speaking to my AI agent and be doing a lot of those things on our behalf. So a long answer. I hope that gives some clarity on what AI agents are and are not some practical use cases, the dangers and what I expect to see in just this next 12 months.

 

Simone Cicero (40:21.912)

Yeah, I mean, I’m fascinated sometimes that since we don’t really understand LLMs, it’s really difficult to, I would say, anticipate their behavior. So one way that we can kind of constrain their behavior is to put some kind of protections, right? So some kind of things that, you know, you cannot go beyond that, something like that, right? And you said that, you know, if you are concerned about your bank account, don’t give it access.

 

But at the same time, I think this is very challenging because on one hand, you want this technology to deal with important things as well, because these are the things that you spend most of your time on. And at the same time, I’m very concerned about letting an LLM kind of reading and sending emails without my control. 

Maybe we can do that if we clearly signal to the recipient of the email that this was written by an AI. So you know that, you know, like when you have your assistant, you know, sending emails for you, it’s not you, it’s your assistant. So maybe you can blame the assistant if things go south.

 

Jeremiah Owyang (41:34.89)

Yeah, but they don’t always disclose that. Most executives have EAs respond on their behalf and you don’t know. You assume it.

 

Simone Cicero (41:41.892)

Yes, but that’s the question for you. You know, you know, would you be comfortable now to open your inbox to an AI and let the AI respond to emails for you? Wouldn’t you be concerned? Yes.

 

Jeremiah Owyang (41:50.954)

Not right now. Definitely not now, not now, but over time. 

 

But don’t forget, Simone, the largest set of data that it can train on is your sent emails. You have probably 10,000 to 100,000 sent emails. Like there’s a lot of data it can train on to look at your behaviors. How did you respond to this person? How did you respond to that person? Who did you not respond to? Who went into spam and filter? Who did you respond to immediately?

 

Like there’s, and which tone did you use? Which conversation did you, how long was it? Was it formal, informal? Was it in Italian? Was it in English? Like there’s already a lot that it can understand to do. And my thought is it wouldn’t, we wouldn’t allow it to do this immediately. We would be guiding and teaching it and reviewing the suggested responses before it hits send. And then over time it will get smarter and smarter and smarter. And then eventually it’ll be seamless. It’ll start responding on our behalf.

 

By the way, I’m not saying this is the best thing in the world. I’m not saying it’s the right thing or the wrong thing. I’m saying it’s a thing it’s going to happen, but the opportunity, the thought is, and, and let’s just go back to my opening remarks here, there’s over 100 AI events in San Francisco in person. Now, why are the AI tech nerds? Why do they want to be in person? Cause they want to be like Europeans and Asians. We actually want to be together. We actually want to be in person, breaking bread, drinking wine, drinking coffee, like talking to each other, interfacing and being human. It is an ironic thing to say, but AI can help us return to our humanity if it does all the labor on our behalf.

 

Simone Cicero (43:37.12)

Yeah, I mean, this opens up a lot of questions on really what is what is labor, how do you get paid, how do you make money in a society like that? Yes, exactly. You know, at the end of the day, what is human? You know, how do you sustain an organization, sorry, a society like that? You know, you know that Marx, you know, was a very adamant in, you know, of course, he didn’t speak about AI. He spoke about something that he used to call general intellect.

 

And essentially the general intellect is like, you know, the knowledge that we create as humans, you know, in society, that can become productive on its own without human intervention at some point. It was so far away in understanding that at some point we could have created an automated intelligence or something like that. And he anticipated that was going to be very critical to the very idea of capitalism. So I guess that we’re gonna, we are in for changes in society and we’ll see what’s going on. So for sure, but I get your points.

 

Jeremiah Owyang (44:38.394)

We do. Actually, that is an important thing. So the concept of universal basic income or heavy taxation and distribution of people who create AI and use AI, like big companies that are going to use AI and reduce employment is a topic that needs to be actually brought to the forefront. I don’t think it’s going to happen with the United States current administration or the next one, based on what I see. So it may not be a discussion that emerges for eight years.

 

I’m sorry, five years, not for five years in the US. It’s already a concept that Northern Europe already knows very well. So there’s models that exist that we need to evaluate. So I’m not opposed to that, even though my title is venture capitalist. We do need to ensure that, hey, I understand. I mean, we work together in the sharing economy. Yes, I always had a, you know, a commerce bent on what I did, but we can’t leave the whole world behind. I totally get that.

 

Simone Cicero (45:37.628)

No, but I think we had similar concerns. You know, when we spoke about the sharing economy 10 years ago, or 12 years ago, we had similar concerns. Shruthi, I know that you have a question coming up.

 

Shruthi Prakash (45:40.245)

Yeah.

 

Shruthi Prakash (45:47.807)

No, I mean, we’ve spoken a lot about what AI can do. So I wanted to know from your side, maybe are there things that it cannot at this stage and if there are misconceptions that it can, but it can’t.

 

Jeremiah Owyang (46:04.974)

That’s a good question. So it can’t hug me like my kids do. It can’t, you know, it can’t make me feel great like my wife does. It cannot play a musical instrument, you know, acoustic instrument with me and look at me and we can improv. I used to play jazz. It can’t do that. And if it does, it’s just not the same. It can’t do those things. It can’t do fitness with me. I go to CrossFit many mornings. It just it’s not the same. I’m with my friends. We’re doing we’re lifting weights. It’s just not the same.

 

Jeremiah Owyang (46:32.95)

So the core human things, it’s not in it and PI and PI and there’s other health apps cannot be equal to a human therapist. It just cannot go to that level and we shouldn’t expect it to. However, not everybody can afford a therapist and they’re not there when you need them at 3 a.m. Not everybody can afford a fitness coach and loneliness is at an all time high. So people want somebody to talk to you at their terms.

 

And this is where AI will come in to help and augment and fill in some of those gaps, never to full capacity. I don’t think ever. Even the most realistic Android, it’s just not enough. So that is where it breaks down that core humanity. Now notice I didn’t say be creative, because I actually think AI can be extremely creative up to the level of the 1% of the creatives breakthrough and break and make a new genre I don’t think AI is going to be able to do that because it’s regurgitating what already existed But it could get up to level 99%

 

Shruthi Prakash (47:43.88)

I mean, I’m curious to see how all of this turns out, right? I’m a psychology student, by the way. So I’m actually really, really curious on that front on how it can sort of imitate that level of, let’s say human innateness, right? Essentially. So I’m curious to see that. So before we come to the end of the discussion, right, I wanted to ask you if you have maybe any breadcrumbs or suggestions for our listeners on books, podcasts, movies that you recommend that they could learn from.

 

Jeremiah Owyang (48:17.678)

Sure. Thank you. So there’s two newsletters that I subscribe to and I know both of the writers and those two ones will give you can stay up to date in the fast changing AI space very quickly with these two. One is called Neuron like the brain neurons and EU are when the other one is called synthedia by Brett. And those two give you all the news and of course they’re using AI to summarize what’s happening.

 

I recommend those two. If you are in Silicon Valley area, Cerebral Valley is the group to watch. And if you’re a developer, they have a great Slack channel. So if you’re building AI, Cerebral Valley, it’s a team. They’re the main group to watch.

 

Simone Cicero (49:05.532)

Mm-hmm. Thank you so much. I mean, I think more than in other episodes, it’s worth that maybe I can do a quick wrap-up of what we discussed, because I think we touched a few very important points during the conversation. And then before closing, maybe you can add some final remark on this. First of all, we spoke about the areas where AI is going to be impactful in your perspective as a VC and as a you know, kind of weaver of networks. And you spoke about productivity, so enterprise use cases, freelance use cases, and so on. Creativity, entertainment, like games and media and so on. And then you spoke about health and wellness. I think these were the things that you highlighted as topics, let’s say. Then we also had a very interesting reflection on business models. And you said software as a service, I guess mostly for enterprise use cases, tokenization for creativity. Then we spoke about media and ads. So it’s very interesting as you paint this future where advertising can be kind of infused into the conversations you have with AI. And I can really think about few use cases for that. And then we spoke about hardware and you said you’re not very much positive towards pins and tabs and things like that. 

 

And maybe as I understand you believe that the phone is maybe the place where the AI will stay for a while. Then we spoke about agents. So really listeners, for you that you are listening, check agents topic and try to understand what is the future because this is the kind of the next wave as I understand well in your perception. And yeah, I mean, of course we discussed all the things, but I think it was worth to pin down this few very important things that kind of provide a framework for people navigate this wave that is coming out. Anything else that you want to add as a final remark on this conversation.

 

Jeremiah Owyang (51:09.426)

The opportunity is that AI is used to help and improve humanity across health and education, which we didn’t talk too much about. Of course, freeing us up from the tasks that we didn’t want to labor on. We saw this in the industrial revolution where people were toiling in the fields and then we shifted. Of course, it has downsides too, but GDP, productivity, more clothes, more food is available to everybody and life spans increased. AI is on that same trajectory.

 

When the industrial revolution, we traded our muscle power for mechanical power, but with AI, and here’s the opportunity, we can trade our thinking power, brain power, to compute power, so everybody in the world, including underdeveloped countries, can benefit, and the whole globe hopefully can improve. I’m excited for the future.

 

Simone Cicero (51:59.084)

Yes, totally. I mean, I can rhyme with this. I was watching a paper a few weeks ago where apparently LLMs are kind of pioneering in medicine and kind of discover the new class of antibiotics. And I was really excited about that. So hopefully, you know, we don’t have all the negative facts, but we also get some positive out of this technology that can really change the world. So thank you so much, Jeremiah, for your enthusiasm, for your pioneering work for being twice a guest on our show. And, you can get back into the last question that we had, the last conversation we had with Jeremiah. It wasn’t in the cusp of the coronavirus pandemic. So it was really rather different ones of the conversation, but I think we went through this and this is also a good signal that we’re gonna go through the challenges that we’re coming up with AI and so on. So thank you so much. I hope you enjoyed the conversation.

 

Jeremiah Owyang (52:59.23)

I absolutely did. Thank you for having me.

 

Simone Cicero (53:02.492)

Thank you, Shruthi. It was always great to have you here and with your questions. Thank you so much.

 

Shruthi Prakash (53:08.547)

Thank you. Thanks Jeremiah for joining. Thank you, Simone.

 

Jeremiah Owyang (53:12.846)

pleasure.

 

Simone Cicero (53:13.172)

And for our listeners, as always, and more than anytime else today, check the show notes. Jeremiah spoke about a few, you know, suggest a few newsletters and tools to use and try. So check the show notes on www.boundareless.io/resources/podcast. You will see that there will be the episode with Jeremiah. And until we speak again, remember to think boundaryless.